Skip to content

[Relax][PyTorch] Fix crash on dynamic shapes with identity slice in ExportedProgram importer #18903

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
tlopex merged 1 commit into from
Mar 10, 2026
Merged

[Relax][PyTorch] Fix crash on dynamic shapes with identity slice in ExportedProgram importer #18903

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion python/tvm/relax/frontend/torch/base_fx_graph_translator.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2076,7 +2076,7 @@ def _reshape(self, node: fx.Node) -> relax.Var:

# Skip identity reshape
current_shape = self.shape_of(x)
if list(current_shape) == list(dims):
if current_shape is not None and list(current_shape) == list(dims):
return x

return self.block_builder.emit(relax.op.reshape(x, dims))
Expand Down
14 changes: 14 additions & 0 deletions python/tvm/relax/frontend/torch/exported_program_translator.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -941,6 +941,20 @@ def _slice(self, node: fx.Node) -> relax.Var:
if end_val is None:
end_val = sys.maxsize

# Skip identity slice (start=0, end>=maxsize, step=1) which is commonly
# emitted by torch.export for dynamic shapes. Without this, strided_slice
# produces shapes like T.min(9223372036854775807, s) that don't simplify,
# causing downstream shape inference failures.
if (
isinstance(start, int)
and isinstance(end_val, int)
and isinstance(step, int)
and start == 0
and end_val >= sys.maxsize
and step == 1
):
Comment on lines +948 to +955

@gemini-code-assist gemini-code-assist Bot Mar 10, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

For improved readability and conciseness, you could group the isinstance checks using the all() built-in function. This would make the condition for identifying an identity slice slightly easier to parse.

        if (
            all(isinstance(v, int) for v in (start, end_val, step))
            and start == 0
            and end_val >= sys.maxsize
            and step == 1
        ):

return x

axes = [dim]
begin = [start]
end = [end_val]
Expand Down
Loading